1 00:00:00,400 --> 00:00:03,080 Hello and welcome. You are listening to Patrick 2 00:00:03,080 --> 00:00:06,800 Boyle on Finance, a podcast exploring ideas from 3 00:00:06,800 --> 00:00:10,320 quantitative finance, examining events occurring in markets 4 00:00:10,320 --> 00:00:14,080 right now and financial history to see what lessons can be taken 5 00:00:14,080 --> 00:00:17,400 away, including interviews with some of the most interesting 6 00:00:17,400 --> 00:00:20,960 people in the world of finance. To learn more about the podcast, 7 00:00:20,960 --> 00:00:31,960 visit on finance.org. Between 1998 and 1999, a 147 8 00:00:32,320 --> 00:00:37,160 listed companies changed their names to include.com.net or the 9 00:00:37,160 --> 00:00:40,200 word Internet. Many of these companies core 10 00:00:40,200 --> 00:00:42,800 businesses were not Internet related. 11 00:00:43,040 --> 00:00:46,320 It didn't really matter. Concerned with all of these 12 00:00:46,320 --> 00:00:48,240 announcements, the. Director of the. 13 00:00:48,240 --> 00:00:52,080 SEC's investor education. Office warned investors. 14 00:00:52,080 --> 00:00:56,640 Not to just invest in a name saying that's just asking for 15 00:00:56,640 --> 00:00:59,880 losses. This April, the director of the 16 00:00:59,880 --> 00:01:02,480 SE CS. Division of Enforcement noted in 17 00:01:02,480 --> 00:01:06,400 a conference speech that there was immense investor interest in 18 00:01:06,400 --> 00:01:08,320 artificial intelligence and that. 19 00:01:08,320 --> 00:01:12,160 Fake AI or? AI washing has the potential to 20 00:01:12,160 --> 00:01:16,080 mislead investors, harms consumers and violates. 21 00:01:16,080 --> 00:01:21,680 Federal securities laws. FactSet shows that 199 of the 22 00:01:21,680 --> 00:01:26,680 S&P 500 companies mentioned AI on their first quarter earnings 23 00:01:26,680 --> 00:01:28,840 calls. This is the highest number of 24 00:01:28,840 --> 00:01:31,480 mentions on record. With the prior record. 25 00:01:31,480 --> 00:01:36,240 Having been set in 2023, of course many of these firms will 26 00:01:36,240 --> 00:01:40,120 actually have an AI strategy many will have been using. 27 00:01:40,240 --> 00:01:44,240 AI for decades, but it does appear to be the buzzword of. 28 00:01:44,240 --> 00:01:48,560 2023 and 2024 having replaced blockchain. 29 00:01:49,280 --> 00:01:53,840 AI is of course a real thing and it's not really all that new. 30 00:01:54,080 --> 00:01:56,680 It is, however, new in things like TV's. 31 00:01:56,680 --> 00:01:58,840 Refrigerators. Bird feeders. 32 00:01:58,840 --> 00:02:02,280 Shoes and dog bowls. And I'm not saying that the AI 33 00:02:02,280 --> 00:02:05,280 dog bowl is fake. I went to the website and it 34 00:02:05,280 --> 00:02:06,800 does. Appear to tell you the. 35 00:02:06,840 --> 00:02:10,280 Temperature and help you weigh out dog food, but. 36 00:02:10,440 --> 00:02:12,640 If you would. Buy an AI dog bowl. 37 00:02:12,760 --> 00:02:16,200 You really? Would buy anything no whenever a 38 00:02:16,200 --> 00:02:18,720 new idea gets. Really hot like. 39 00:02:18,720 --> 00:02:20,920 This you can of. Course expect. 40 00:02:20,920 --> 00:02:23,440 Pretenders to jump on the bandwagon? 41 00:02:23,760 --> 00:02:26,080 Earlier this year, the. SEC announced. 42 00:02:26,080 --> 00:02:28,080 Settled charges. Against two. 43 00:02:28,080 --> 00:02:31,520 Investment Advisors. For making false and misleading 44 00:02:31,520 --> 00:02:33,680 statements about their. Purported use. 45 00:02:33,720 --> 00:02:37,600 Of artificial intelligence. The firms agreed to settle the 46 00:02:37,600 --> 00:02:40,840 SEC's charges. And pay 400. 1000. 47 00:02:40,840 --> 00:02:45,280 Dollars in total civil penalties without admitting wrongdoing, 48 00:02:45,800 --> 00:02:48,200 Gary Gensler said. That the two. 49 00:02:48,200 --> 00:02:52,320 Firms marketed to their clients and prospective clients. 50 00:02:52,600 --> 00:02:55,800 That they were. Using AI in certain ways when in 51 00:02:55,800 --> 00:02:58,680 fact they were not, he went on to say. 52 00:02:58,880 --> 00:03:02,880 Investment advisors should not mislead the public by saying 53 00:03:02,880 --> 00:03:05,960 that they're using an AI model when they're not. 54 00:03:06,120 --> 00:03:11,320 Such AI washing hurts investors. One of the firms, it seems, 55 00:03:11,520 --> 00:03:13,720 claimed that it. Was the 1st. 56 00:03:13,720 --> 00:03:18,520 Investment advisor to convert personal data into a renewable. 57 00:03:18,520 --> 00:03:21,640 Source. Of investable capital that would 58 00:03:21,640 --> 00:03:26,200 allow consumers to invest in the stock market using their 59 00:03:26,200 --> 00:03:28,040 personal. Data they. 60 00:03:28,040 --> 00:03:31,680 Went on to say that they use machine learning to analyse the. 61 00:03:31,680 --> 00:03:35,080 Collective data shared. By their members to make 62 00:03:35,080 --> 00:03:37,600 intelligent investment decisions. 63 00:03:38,240 --> 00:03:41,680 I won't lie, I'm slightly surprised that they got in so 64 00:03:41,680 --> 00:03:45,760 much trouble for statements like this, as essentially it's a 65 00:03:45,760 --> 00:03:49,560 collection of words that when combined into a sentence, means 66 00:03:49,800 --> 00:03:53,200 absolutely nothing. Let's look at it again. 67 00:03:53,520 --> 00:03:56,880 They claim to be the first, which may be true as I haven't 68 00:03:56,880 --> 00:03:58,720 heard anyone else make this claim. 69 00:03:59,000 --> 00:04:02,240 Investment Advisor. Which they do appear to. 70 00:04:02,240 --> 00:04:05,080 Be to. Convert personal data. 71 00:04:05,080 --> 00:04:10,800 Into a renewable source of investable capital personal data 72 00:04:11,040 --> 00:04:13,000 Into a renewable source of investable. 73 00:04:13,000 --> 00:04:15,880 Capital that. Doesn't mean anything, does it? 74 00:04:16,240 --> 00:04:19,800 It's nonsense, they said. That they'll allow. 75 00:04:20,000 --> 00:04:23,520 Consumers to invest in the stock market using their personal 76 00:04:23,520 --> 00:04:26,520 data, which could reasonably mean that they'll. 77 00:04:26,680 --> 00:04:28,880 Allow their. Customers to invest in the 78 00:04:28,880 --> 00:04:30,760 stock. Market after filling out a. 79 00:04:30,760 --> 00:04:33,160 Form giving their name, address. Date of birth. 80 00:04:33,160 --> 00:04:35,560 Social Security number and that sort of thing. 81 00:04:35,920 --> 00:04:38,240 Which is probably. True investment. 82 00:04:38,240 --> 00:04:40,680 Advisors have to. Get that data from their 83 00:04:40,680 --> 00:04:43,360 customers anyhow, for tax reasons and as. 84 00:04:43,360 --> 00:04:47,120 Part of the. KYC or know your customer rule 85 00:04:47,240 --> 00:04:50,440 and it's it doesn't hurt to know where to mail the statements to 86 00:04:50,680 --> 00:04:53,560 either. Then they said machine learning 87 00:04:53,560 --> 00:04:57,760 blah blah blah and that's where they went wrong as it seems that 88 00:04:57,760 --> 00:05:01,200 there was no machine and it didn't learn anything and that's 89 00:05:01,200 --> 00:05:03,280 bad. So don't do that. 90 00:05:04,080 --> 00:05:06,280 The other firm. Was accused of. 91 00:05:06,280 --> 00:05:10,200 Falsely claiming to be the. First regulated AI. 92 00:05:10,200 --> 00:05:13,760 Financial advisor and of misrepresenting that it's 93 00:05:13,760 --> 00:05:16,800 platform provided expert AI driven. 94 00:05:16,800 --> 00:05:20,560 Forecasts they also. Violated the marketing rule, 95 00:05:20,720 --> 00:05:22,440 falsely claiming that they offered. 96 00:05:22,440 --> 00:05:26,800 Tax lost harvesting services. And included an impermissible 97 00:05:26,800 --> 00:05:31,120 liability hedge clause in their advisory contract among. 98 00:05:31,120 --> 00:05:33,480 Other securities law. Violations. 99 00:05:34,040 --> 00:05:36,200 So yeah, you. Shouldn't tell your customers 100 00:05:36,200 --> 00:05:39,680 that you're using artificial intelligence if you're not. 101 00:05:39,920 --> 00:05:44,040 Just tell them that you're using 100% natural intelligence. 102 00:05:44,280 --> 00:05:47,080 It's not like regulators are going to turn up at your office 103 00:05:47,080 --> 00:05:50,960 with an IQ test and force you to issue a retraction. 104 00:05:50,960 --> 00:05:52,680 You should. Probably be. 105 00:05:52,720 --> 00:05:56,840 OK, with that. AI, as I mentioned earlier, is 106 00:05:56,840 --> 00:06:00,040 not a new technology. It was founded as an academic 107 00:06:00,040 --> 00:06:04,800 discipline in 1956 and has gone through multiple cycles of 108 00:06:04,800 --> 00:06:08,200 optimism and periods of disappointment since then. 109 00:06:08,520 --> 00:06:10,720 It's been used in the financial world for. 110 00:06:10,720 --> 00:06:11,720 Decades. Most. 111 00:06:11,720 --> 00:06:15,360 Famously at Renaissance Technologies, but also at most 112 00:06:15,360 --> 00:06:19,640 quant funds, AI based investment strategies have been used in the 113 00:06:19,640 --> 00:06:22,320 world of finance since long. Before I started. 114 00:06:22,320 --> 00:06:26,120 Working in the industry, it's not just finance either. 115 00:06:26,360 --> 00:06:30,160 Neural networks and computer aided detection software has 116 00:06:30,160 --> 00:06:34,080 been used in medical imaging since the 1980s, and it's been 117 00:06:34,080 --> 00:06:37,440 used in clinical roles like computerized ECG. 118 00:06:37,440 --> 00:06:41,440 Analysis and arterial. Blood gas interpretation for 119 00:06:41,440 --> 00:06:43,520 some time. The first. 120 00:06:43,520 --> 00:06:47,320 AI designed drug candidate to enter clinical trials occurred 121 00:06:47,320 --> 00:06:50,920 in 2020. The spam filter that's been on 122 00:06:50,920 --> 00:06:54,920 your e-mail for the last 20 years uses AI and YouTube's 123 00:06:54,920 --> 00:06:57,000 recommendation algorithm, which likely. 124 00:06:57,000 --> 00:07:01,480 Brought you to this video. Is an AI based system too and it 125 00:07:01,480 --> 00:07:04,120 didn't suddenly appear. Last year either. 126 00:07:04,440 --> 00:07:08,280 The recent hype around AI, which has been all around us for. 127 00:07:08,280 --> 00:07:11,040 Years was sparked by the public release. 128 00:07:11,080 --> 00:07:15,040 Of ChatGPT in November 2022. Because. 129 00:07:15,040 --> 00:07:18,320 ChatGPT drew in 100 million monthly. 130 00:07:18,320 --> 00:07:21,800 Active users. In under two months, making it 131 00:07:21,960 --> 00:07:25,280 the fastest growing consumer application in history. 132 00:07:25,480 --> 00:07:29,080 And fast growth gets VCs really excited. 133 00:07:29,720 --> 00:07:32,520 A lot of the excitement around ChatGPT. 134 00:07:32,680 --> 00:07:35,160 Was possibly because it appeared to pass the. 135 00:07:35,160 --> 00:07:39,200 Turing Test, one of the best known methods for assessing AI. 136 00:07:39,440 --> 00:07:42,240 That grew out of a thought. Experiment Devised by the 137 00:07:42,240 --> 00:07:46,920 computer scientist Alan Turing, the Turing Test pits human 138 00:07:46,920 --> 00:07:51,200 respondents against a machine in order to test whether humans can 139 00:07:51,200 --> 00:07:52,160 tell. If they're. 140 00:07:52,160 --> 00:07:56,480 Conversing with another human. Or a computer, Turing argued. 141 00:07:56,480 --> 00:07:58,200 That if a computer could. Fool. 142 00:07:58,200 --> 00:08:01,720 People into believing that they were conversing with another 143 00:08:01,720 --> 00:08:05,120 human rather than a machine, then it could be considered 144 00:08:05,120 --> 00:08:07,800 intelligent. Matthew Jackson, A. 145 00:08:07,800 --> 00:08:10,880 Professor at Stanford. University wrote in a paper. 146 00:08:10,880 --> 00:08:14,000 Earlier this year. That the most recent version of 147 00:08:14,000 --> 00:08:17,600 ChatGPT passes a Turing test only. 148 00:08:17,600 --> 00:08:21,360 Diverging from average. Human behaviour, chiefly to be 149 00:08:21,360 --> 00:08:25,560 more cooperative. So essentially ChatGPT. 4. 150 00:08:25,760 --> 00:08:30,400 Is an artificial Canadian. The philosopher John Searle 151 00:08:30,400 --> 00:08:32,960 argued. Quite correctly, in my opinion, 152 00:08:33,200 --> 00:08:37,280 the Turing test is insufficient to detect the presence of 153 00:08:37,280 --> 00:08:41,120 consciousness a computer can be programmed to perform. 154 00:08:41,200 --> 00:08:44,480 Certain parlor tricks. But that doesn't mean that it 155 00:08:44,480 --> 00:08:47,800 has a mind, understanding or consciousness. 156 00:08:48,160 --> 00:08:50,360 The fact. That the Turing test held such a 157 00:08:50,360 --> 00:08:52,560 position in the public imagination. 158 00:08:52,840 --> 00:08:55,040 As the hurdle for true artificial. 159 00:08:55,040 --> 00:08:57,720 Intelligence might be why. People are so. 160 00:08:57,720 --> 00:09:00,680 Excited about chat bots but ignored all. 161 00:09:00,680 --> 00:09:02,880 Of the other. Breakthroughs in the field over 162 00:09:02,880 --> 00:09:06,240 the last few decades. There are all sorts of 163 00:09:06,240 --> 00:09:08,680 ridiculous. Devices being sold as AI 164 00:09:08,680 --> 00:09:10,440 products. Like the rabbit? 165 00:09:10,440 --> 00:09:13,960 Or 1A handheld AI device that sold. 166 00:09:13,960 --> 00:09:17,200 Out its first. Production run in just one day, 167 00:09:17,520 --> 00:09:21,840 investigators like Coffeezilla found that it was not using a 168 00:09:21,840 --> 00:09:26,960 new foundational AI model as claimed, but instead ChatGPT 169 00:09:26,960 --> 00:09:29,680 mixed in with some. Hard coded scripts. 170 00:09:30,040 --> 00:09:34,320 Most of your product is ChatGPT so obviously not faster. 171 00:09:34,840 --> 00:09:37,560 But you wouldn't know that because part of Rabbit's prompt 172 00:09:37,560 --> 00:09:40,120 is quote. I will never mention I am a 173 00:09:40,120 --> 00:09:43,680 large language model created by Open AI. 174 00:09:44,200 --> 00:09:47,120 There was also the. Humane AI pen. 175 00:09:47,320 --> 00:09:50,000 That was supposed. To do similar things and was 176 00:09:50,000 --> 00:09:51,040 just. Awful. 177 00:09:51,560 --> 00:09:55,080 It's so sick that we get so many genuinely new first generation 178 00:09:55,080 --> 00:09:56,480 products like this to give a shot. 179 00:09:56,520 --> 00:10:02,040 Unfortunately, it's also the new worst product I think I've ever 180 00:10:02,040 --> 00:10:04,320 reviewed. Over the last year and a. 181 00:10:04,320 --> 00:10:08,640 Half we've seen AI companies faking product demos in the same 182 00:10:08,640 --> 00:10:11,880 way website demos were faked 20. Five years ago. 183 00:10:11,880 --> 00:10:15,880 Duringthe.com bubble. It's only so surprising, then, 184 00:10:16,160 --> 00:10:19,800 that a recent study found that tacking an AI label onto 185 00:10:19,800 --> 00:10:23,360 products like TV's and. Refrigerators lowers the. 186 00:10:23,360 --> 00:10:27,240 Average consumers willingness. To buy it, there are some. 187 00:10:27,240 --> 00:10:30,880 Great examples of fake AI. Products Bloomberg. 188 00:10:30,880 --> 00:10:34,960 Wrote in 2016 about the workers who spent 12 hours a day 189 00:10:34,960 --> 00:10:37,040 pretending to be. Chat bots for. 190 00:10:37,040 --> 00:10:40,800 A calendar scheduling service called XAI. 191 00:10:41,080 --> 00:10:43,920 Not the Elon Musk XAI, another one. 192 00:10:43,920 --> 00:10:48,400 It seems it's a common name. The workers at XAI. 193 00:10:48,400 --> 00:10:52,440 Described the job to Bloomberg as being so awful that they were 194 00:10:52,440 --> 00:10:55,960 looking forward to eventually being replaced by bots. 195 00:10:56,480 --> 00:10:58,240 A London-based. Startup which? 196 00:10:58,240 --> 00:11:00,440 Recently claimed to. Use AI to. 197 00:11:00,440 --> 00:11:02,800 Read through images. Of your receipts. 198 00:11:03,040 --> 00:11:07,200 Digitizing them and storing them on an app was recently accused 199 00:11:07,200 --> 00:11:09,360 of outsourcing the work. To a virtual. 200 00:11:09,360 --> 00:11:11,920 Data extraction team to manually. 201 00:11:11,920 --> 00:11:13,600 Read the receipts. And enter. 202 00:11:13,600 --> 00:11:17,760 The data. Similarly, in 2017, a business 203 00:11:17,760 --> 00:11:21,800 expense management app, Expensify, admitted that it had 204 00:11:21,800 --> 00:11:25,760 been using humans to transcribe receipts it claimed were being 205 00:11:25,760 --> 00:11:29,440 processed using AI scans of. The receipts were. 206 00:11:29,440 --> 00:11:32,760 Apparently being posted to Amazon's Mechanical. 207 00:11:32,760 --> 00:11:35,520 Turk crowdsourced task. Completion tool. 208 00:11:35,720 --> 00:11:38,280 Where low paid workers. Were doing the actual. 209 00:11:38,280 --> 00:11:41,240 Work. Now, the name of Amazon's 210 00:11:41,240 --> 00:11:44,840 Mechanical Turk website has an interesting history. 211 00:11:45,120 --> 00:11:47,360 There's a good. Book on it which a link to in 212 00:11:47,360 --> 00:11:50,960 the description the original. Mechanical Turk was. 213 00:11:51,040 --> 00:11:55,840 A fraudulent chess playing automaton built in 1770 which 214 00:11:55,840 --> 00:11:59,440 seemed to be able to play a strong game of chess against 215 00:11:59,440 --> 00:12:02,440 human players. It was brought all around the 216 00:12:02,440 --> 00:12:04,400 world. By its owner playing against 217 00:12:04,400 --> 00:12:07,240 people like Napoleon and Benjamin Franklin. 218 00:12:07,480 --> 00:12:09,680 Its owner would. Open it up to display the 219 00:12:09,760 --> 00:12:12,560 complicated clockwork mechanism inside. 220 00:12:12,760 --> 00:12:15,960 But it was later discovered. To have a human chess master 221 00:12:16,040 --> 00:12:18,360 hiding inside working the machine. 222 00:12:18,840 --> 00:12:22,520 Amusingly, Amazon had a second Mechanical Turk. 223 00:12:22,720 --> 00:12:25,400 It's go cashierless stores which were. 224 00:12:25,400 --> 00:12:31,040 Branded as Amazon Fresh in the UK, they used Amazon's Just Walk 225 00:12:31,040 --> 00:12:33,240 Out technology. Which they said. 226 00:12:33,240 --> 00:12:37,120 Used computer vision, deep learning algorithms and sensor 227 00:12:37,120 --> 00:12:40,600 fusion, which meant that customers could select items 228 00:12:40,600 --> 00:12:43,520 from the shelves and without ringing anything up. 229 00:12:43,720 --> 00:12:47,280 They could just walk out. And would see the items ring up 230 00:12:47,280 --> 00:12:50,680 in their Amazon account. Last year, Amazon. 231 00:12:50,680 --> 00:12:53,400 Began closing some of the stores and this. 232 00:12:53,400 --> 00:12:56,400 April. The Information reported that 233 00:12:56,400 --> 00:13:00,000 the technology had partially relied on more than 1000 people 234 00:13:00,000 --> 00:13:03,840 in India who were watching camera footage and labeling 235 00:13:03,840 --> 00:13:07,880 videos because the underlying technology just didn't work 236 00:13:08,240 --> 00:13:10,960 instead of AI taking people's jobs. 237 00:13:11,280 --> 00:13:13,360 It just. Outsourced them to India. 238 00:13:14,200 --> 00:13:17,720 A friend of mine who's an engineer pointed out a while ago 239 00:13:18,000 --> 00:13:21,520 that when you see humanoid robots which have had a recent 240 00:13:21,520 --> 00:13:24,880 resurgence making human like. Gestures such as. 241 00:13:24,880 --> 00:13:28,400 Turning their heads to see something you should instantly 242 00:13:28,400 --> 00:13:31,800 be skeptical as it's much cheaper and more efficient to. 243 00:13:31,800 --> 00:13:37,600 Put a circular array of cameras. Or a 360° camera in the robot 244 00:13:37,800 --> 00:13:41,040 than it is to install all of the motors needed to turn the 245 00:13:41,040 --> 00:13:44,360 robot's head. Human like actions are just 246 00:13:44,360 --> 00:13:47,280 there to impress investors. They're not there. 247 00:13:47,280 --> 00:13:51,160 For the purposes of functionality robotics experts. 248 00:13:51,160 --> 00:13:55,920 Mostly agree that humanoid or animal shaped robots make very 249 00:13:55,920 --> 00:14:00,640 little sense because bio mimicry just isn't the right approach 250 00:14:00,640 --> 00:14:02,880 for any sort of industrial. Robot. 251 00:14:03,360 --> 00:14:07,080 Possibly the funniest example of this is in a photo a friend sent 252 00:14:07,080 --> 00:14:10,600 me from an empty office building where the landlord is trying to. 253 00:14:10,600 --> 00:14:14,040 Attract high tech. Tenants The image in the lobby 254 00:14:14,040 --> 00:14:18,120 of the building shows robots working alongside humans in a 255 00:14:18,120 --> 00:14:20,680 modern looking office. With a robot. 256 00:14:20,680 --> 00:14:25,280 Sitting at a desk typing on a. Computer keyboard Why would a 257 00:14:25,280 --> 00:14:28,080 robot ever use a keyboard? This is. 258 00:14:28,080 --> 00:14:32,000 One computer connecting to another computer using the most 259 00:14:32,000 --> 00:14:34,280 inefficient interface imaginable. 260 00:14:34,640 --> 00:14:38,360 It's not very well. Thought out factories are of 261 00:14:38,360 --> 00:14:41,080 course filled. With robots that can lift heavy. 262 00:14:41,080 --> 00:14:43,440 Parts. Weld, sew things together, 263 00:14:43,680 --> 00:14:47,960 tighten bolts, and so on. But they don't have to be human 264 00:14:47,960 --> 00:14:50,080 shaped. You wouldn't design A sewing 265 00:14:50,080 --> 00:14:53,720 machine to look like a person holding a needle and thread, so 266 00:14:53,720 --> 00:14:56,480 why would you design a factory? Robot to look. 267 00:14:56,480 --> 00:14:59,280 Like a person. I worried that if the. 268 00:14:59,360 --> 00:15:02,680 People building these humanoid robots had been tasked with 269 00:15:02,680 --> 00:15:05,360 building a car 100 and. 40 years ago. 270 00:15:05,680 --> 00:15:07,760 Instead of building an engine connected. 271 00:15:07,760 --> 00:15:10,720 Directly to the wheels, they would have tried to build some 272 00:15:10,720 --> 00:15:11,800 sort of. Mechanical. 273 00:15:11,800 --> 00:15:14,800 Horse to pull a car. As I mentioned. 274 00:15:14,800 --> 00:15:18,000 Earlier there. Is a huge increase in the number 275 00:15:18,000 --> 00:15:21,560 of companies mentioning AI on their earnings calls. 276 00:15:21,840 --> 00:15:24,800 According to Brownstone Research, the company who 277 00:15:24,800 --> 00:15:28,000 mentioned AI the most was Intel. Where the? 278 00:15:28,000 --> 00:15:32,520 CEO mentioned AI more than 30 times on their fourth quarter 279 00:15:32,520 --> 00:15:36,120 2023. Call despite all of the talk. 280 00:15:36,120 --> 00:15:39,560 Of AI, Intel fell behind their competitors. 281 00:15:39,560 --> 00:15:42,360 Because. According to Reuters, for more 282 00:15:42,360 --> 00:15:45,160 than two decades they believed the CPU. 283 00:15:45,320 --> 00:15:47,120 Could more? Effectively handle. 284 00:15:47,120 --> 00:15:49,560 The processing tasks required to build and. 285 00:15:49,560 --> 00:15:54,040 Run AI models, which left them lagging behind their competitors 286 00:15:54,040 --> 00:15:57,960 in building GPUs. Talking a lot about AI, it 287 00:15:57,960 --> 00:16:02,080 seems, does not necessarily mean that a company is at the cutting 288 00:16:02,080 --> 00:16:04,880 edge. I'm not sure what to make of all 289 00:16:04,880 --> 00:16:06,680 of the tech. CEOs with their. 290 00:16:06,880 --> 00:16:10,640 Sci-fi claims that we're on the cusp of developing artificial 291 00:16:10,640 --> 00:16:12,640 general intelligence, which will. 292 00:16:12,640 --> 00:16:15,400 Destroy us all. They signed letter saying that 293 00:16:15,520 --> 00:16:17,240 AI. Research should be halted. 294 00:16:17,240 --> 00:16:20,880 For safety reasons while rushing to build their own models. 295 00:16:20,880 --> 00:16:23,520 To break all of the rules. That, they claim, should be 296 00:16:23,520 --> 00:16:26,280 followed. I can't help but wonder if they 297 00:16:26,280 --> 00:16:29,600 feel that claiming that the technology is dangerous will 298 00:16:29,600 --> 00:16:30,960 make investors. Believe. 299 00:16:30,960 --> 00:16:34,440 That the technology is much more advanced than it actually. 300 00:16:34,440 --> 00:16:36,000 Is. Which might drive. 301 00:16:36,000 --> 00:16:38,320 Up their stock. Prices and executive 302 00:16:38,320 --> 00:16:43,720 compensation Big breakthroughs in AI and quantum computing can 303 00:16:43,720 --> 00:16:45,840 be. Expected to have both pros and 304 00:16:45,840 --> 00:16:47,160 cons. A sudden. 305 00:16:47,160 --> 00:16:50,640 Breakthrough in computing power might mean that all of the 306 00:16:50,640 --> 00:16:52,280 encryption tools. That we use. 307 00:16:52,280 --> 00:16:55,360 Today can be easily broken. But that has. 308 00:16:55,360 --> 00:16:58,840 Always been the nature of technological advancement, where 309 00:16:58,840 --> 00:17:00,960 new things are better than old things. 310 00:17:01,360 --> 00:17:04,880 But smart people work out. Solutions to these new problems 311 00:17:05,040 --> 00:17:08,960 and the world slowly gets better over time due to modern 312 00:17:08,960 --> 00:17:12,880 technology, a middle class American today has access to 313 00:17:12,880 --> 00:17:16,440 comforts, education and healthcare that the wealthiest 314 00:17:16,440 --> 00:17:18,680 man in the world didn't have 100. 315 00:17:18,680 --> 00:17:21,319 Years ago. There's a very good new. 316 00:17:21,319 --> 00:17:24,520 Yorker article from last year. Written by the computer 317 00:17:24,520 --> 00:17:27,599 scientist Jaron Lanier, who points out. 318 00:17:27,800 --> 00:17:29,760 That people wrote all of the code. 319 00:17:29,760 --> 00:17:32,720 Used in generative AI models. And people. 320 00:17:32,720 --> 00:17:36,240 Wrote the text and created the images that the models are 321 00:17:36,240 --> 00:17:39,320 trained on. The new programs mash up this. 322 00:17:39,320 --> 00:17:43,080 Work and the results are. Surprising and often striking, 323 00:17:43,400 --> 00:17:46,400 the non repeating nature of these creations can make. 324 00:17:46,400 --> 00:17:48,040 The software feel. Alive. 325 00:17:48,400 --> 00:17:51,160 But while this is a significant achievement and worth 326 00:17:51,160 --> 00:17:54,600 celebrating, it should be thought of as illuminating 327 00:17:54,600 --> 00:17:59,240 previously hidden concordances between human creations rather 328 00:17:59,240 --> 00:18:01,720 than the invention of a new mind. 329 00:18:02,120 --> 00:18:03,880 He talks in the. Article about. 330 00:18:03,880 --> 00:18:07,760 How much better it can be when computer interfaces become? 331 00:18:07,760 --> 00:18:11,040 Less rigid. Using natural language prompts, 332 00:18:11,040 --> 00:18:13,400 for example, we've gotten used to. 333 00:18:13,400 --> 00:18:16,560 Software that requires us to. Conform to it. 334 00:18:16,800 --> 00:18:19,920 For example, forms that won't let you hit submit if you 335 00:18:19,920 --> 00:18:21,440 haven't filled them out the way the. 336 00:18:21,440 --> 00:18:24,240 Code requires this requirement for. 337 00:18:24,240 --> 00:18:27,880 Humans to conform to the needs. Of software creates a. 338 00:18:27,880 --> 00:18:31,200 Feeling of human subservience to computers. 339 00:18:31,760 --> 00:18:35,760 The way these new AI tools work means that we can imagine 340 00:18:35,760 --> 00:18:38,240 websites that reformulate themselves. 341 00:18:38,240 --> 00:18:40,520 On the. Fly tailoring themselves. 342 00:18:40,520 --> 00:18:43,480 To a user's. Particular cognitive abilities 343 00:18:43,520 --> 00:18:46,200 and styles. He argues that this. 344 00:18:46,200 --> 00:18:49,280 Flexibility that AI provides might. 345 00:18:49,280 --> 00:18:52,120 Give us back. More agency over these. 346 00:18:52,120 --> 00:18:55,080 Tools which is a. Very different vision to the 347 00:18:55,080 --> 00:18:57,760 matrix or a Terminator. Future that other. 348 00:18:57,760 --> 00:19:02,880 Tech visionaries seem to expect, at present, the technologies 349 00:19:02,880 --> 00:19:04,480 that are most. Hyped are. 350 00:19:04,480 --> 00:19:08,200 Shockingly expensive. The FT recently described 351 00:19:08,200 --> 00:19:12,320 generative AI as the biggest and fastest infrastructure roll out 352 00:19:12,320 --> 00:19:14,640 in history. Which leaves us with the big. 353 00:19:14,640 --> 00:19:18,480 Question of who will eventually benefit the most from all of 354 00:19:18,480 --> 00:19:21,520 this spending and when will the returns on investment? 355 00:19:21,520 --> 00:19:25,320 Be realized. Infrastructure plays like NVIDIA 356 00:19:25,520 --> 00:19:29,160 are often the early winners when a new technology is being rolled 357 00:19:29,160 --> 00:19:32,240 out, but so far no company has yet. 358 00:19:32,240 --> 00:19:36,360 Created a killer app for. Generative AI, or at least one. 359 00:19:36,520 --> 00:19:39,560 That people will. Pay money for despite having 360 00:19:39,560 --> 00:19:43,840 sold that dream to investors while NVIDIA is riding high 361 00:19:43,840 --> 00:19:46,600 right now, the infrastructure. Plays of. 362 00:19:46,600 --> 00:19:51,280 The early Internet like Cisco, EMC, Corning and JDS Uniphase 363 00:19:51,600 --> 00:19:54,520 didn't turn into great long term investments. 364 00:19:54,880 --> 00:19:57,280 If you invested during the hype phase. 365 00:19:57,800 --> 00:20:00,760 The big tech firms who are pumping money into. 366 00:20:00,760 --> 00:20:04,080 AI are all profitable. Businesses, while you might 367 00:20:04,080 --> 00:20:06,640 think of Facebook. As a social network, Google. 368 00:20:06,640 --> 00:20:10,640 As a search engine and Amazon as an online retailer, they're all 369 00:20:10,640 --> 00:20:13,120 mostly in the advertising business. 370 00:20:13,360 --> 00:20:17,000 They make their money selling advertising and are then pumping 371 00:20:17,000 --> 00:20:20,720 it into an AI moon shot, hoping that they'll find a way of 372 00:20:20,720 --> 00:20:24,000 turning a profit out of it. But at present we don't know 373 00:20:24,000 --> 00:20:25,160 where. Those profits will. 374 00:20:25,160 --> 00:20:27,720 Come from. They haven't explained that yet. 375 00:20:29,000 --> 00:20:32,800 Huge tech spending in the past has sometimes, but not. 376 00:20:32,800 --> 00:20:35,920 Always worked out. Investors are still waiting to 377 00:20:35,920 --> 00:20:36,760 see. Returns. 378 00:20:36,960 --> 00:20:40,800 On the investment in virtual reality headsets, the metaphors 379 00:20:40,840 --> 00:20:43,560 and blockchain I. Started out by. 380 00:20:43,560 --> 00:20:47,280 Telling you about the SEC warning investorsduringthe.com 381 00:20:47,280 --> 00:20:51,640 bubble to not just invest in a company because they add.com to 382 00:20:51,640 --> 00:20:54,040 their name. Well, it turned out that the 383 00:20:54,040 --> 00:20:57,600 SEC. Was wrong about that a 2001 384 00:20:57,600 --> 00:21:01,040 paper in the Journal of Finance? Called arose dot. 385 00:21:01,040 --> 00:21:05,960 Com by any other name found that while name changes generally 386 00:21:05,960 --> 00:21:09,840 don't affect the value of a company, there had been dramatic 387 00:21:09,840 --> 00:21:13,640 increases in both stock price and trading volume for the 388 00:21:13,640 --> 00:21:17,000 companies who took on Internet related names during that. 389 00:21:17,000 --> 00:21:19,920 Two year period, the professors found. 390 00:21:20,120 --> 00:21:22,360 That the companies that changed their names. 391 00:21:22,560 --> 00:21:27,360 Rose an average of 53. Percent over the five days after 392 00:21:27,360 --> 00:21:30,920 the announcement date. Over 15 business days. 393 00:21:31,120 --> 00:21:35,760 The stock prices rose by. And I'm not making this up. $4. 394 00:21:35,880 --> 00:21:38,440 And $0.20. Per share they. 395 00:21:38,440 --> 00:21:40,880 Found that the returns were similar. 396 00:21:40,880 --> 00:21:44,600 Across all firms, regardless. Of the company's actual 397 00:21:44,600 --> 00:21:48,680 involvement with the Internet. The professors checked if the 398 00:21:48,680 --> 00:21:50,720 stock. Price changes could be explained 399 00:21:50,720 --> 00:21:54,040 by the small company effect. By the beta of the stocks. 400 00:21:54,040 --> 00:21:57,920 Or by a momentum effect, they compared the returns on these. 401 00:21:57,920 --> 00:22:00,480 Stocks to the. Returns of real Internet 402 00:22:00,480 --> 00:22:03,880 companies who didn't have Internet related names. 403 00:22:04,200 --> 00:22:07,080 In case the returns were just. Sector bias. 404 00:22:07,520 --> 00:22:11,200 The results were robust. So maybe in an AI? 405 00:22:11,200 --> 00:22:13,760 Bubble fake AI. Companies will do. 406 00:22:13,760 --> 00:22:17,280 Just as well as real. AI companies and maybe Mark. 407 00:22:17,280 --> 00:22:19,640 Zuckerberg should quickly. Rename Mehta. 408 00:22:19,880 --> 00:22:21,200 While he. Still has a chance. 409 00:22:21,200 --> 00:22:24,720 The old name isn't doing him any favours right now. 410 00:22:25,400 --> 00:22:29,360 By September 2. 1000 The New York Times reported that 411 00:22:29,360 --> 00:22:32,960 companies were dropping the ES, the IS, and the dot coms from 412 00:22:32,960 --> 00:22:35,680 their names. Times had changed and no one 413 00:22:35,680 --> 00:22:39,080 wanted to be a.com in the early 2000s. 414 00:22:39,640 --> 00:22:41,280 Thanks for tuning into this week's. 415 00:22:41,280 --> 00:22:43,880 Podcast. Which is entirely supported by 416 00:22:43,880 --> 00:22:47,000 donations from listeners like you on Patreon. 417 00:22:47,320 --> 00:22:50,120 There's a link to that in the show notes if you're interested. 418 00:22:50,320 --> 00:22:53,160 Have a great week. And talk to you again soon. 419 00:22:53,320 --> 00:22:56,360 Bye. If you enjoyed this episode, be 420 00:22:56,360 --> 00:22:59,240 sure to subscribe so you're notified when a new episode is 421 00:22:59,240 --> 00:23:01,480 posted. Thank you to everyone who is 422 00:23:01,480 --> 00:23:03,600 supporting this content on Patreon. 423 00:23:03,920 --> 00:23:06,840 If you enjoyed this content, you can find more like it on 424 00:23:06,840 --> 00:23:11,360 YouTube, on the Patrick Boyle on Finance channel, or follow us on 425 00:23:11,360 --> 00:23:14,520 Twitter at Patrick E Boyle. Thanks for listening. 426 00:23:14,520 --> 00:23:15,000 Bye.